A Comparative Study of Machine Learning Algorithms Using Explainable Artificial Intelligence System for Predicting Liver Disease

Abstract

Artificial Intelligence (AI) has become increasingly integral to healthcare, aiding medical professionals in disease diagnosis. However, a critical challenge faced by practitioners lies in the trustworthiness of diagnosed results, as machine learning models often lack transparent explanations for their predictions. Explainable AI (XAI) addresses this, allowing practitioners and patients to comprehend and trust outcomes. This study employs the liver disease patient dataset to predict liver disease, vital in India where yearly, 10 lakh people face new liver cirrhosis diagnoses with high mortality. Early detection difficulties underscore the need for transparent predictions. Using Kaggle data, Random Forest, XGBoost, and Explainable Boosting Machine (EBM) classifiers are compared, with EBM outshining, achieving a 99.8% accuracy score. The study concludes that the EBM classifier is a superior and transparent liver disease predictor.